SQL count by group

Question

I have to following schema

Movie(mvID, title, rating, year)
Director(directorID, firstname, lastname)
Genre(mvID*, genre)
Direct(mvID*, directorID*)

Answer 1

The SQL below will solve your problem:

SELECT      Director.DirectorID, Director.FirstName, Director.LastName, COUNT(*)
FROM        Direct, Genre, Director
WHERE       Direct.mvID = Genre.mvID
AND     Genre.genre = 'Comedy'
AND     Direct.DirectorID = Director.DirectorID
AND     rownum = 1
GROUP BY    Director.DirectorID, Director.FirstName, Director.LastName
order by    COUNT(*) DESC

Specifically, this joins the tables and counts the movies directed by each director for movies of genre 'Comedy'. Then it sorts the row in descending order of count and grabs the first row.

Note you have two many<->many relationships so if you weren't only looking at comedy (i.e. if you ran this across multiple genres), you would probably have to use a different technique (i.e. multiple temp tables or similar virtual aggregates) in order to not have 'double counting in your SQL...

Also note that if two directors have the same count of movies for this genre, only one row would be brought back. You would have to modify the sql slightly if you wanted all directors with the same count of comedies to be returned.

Cheers, Noah Wollowick

Answer 2

select d.firstname,d.lastname,count(g.mvId)
from Director d
inner join Direct dr on d.directorId=dr.directorId
inner join Genre g on dr.mvId=g.mvId
where g.genre='comedy'
group by d.firstname
having count(g.mvId)=max(g.mvId)